Intel i7 TLB bug

[MarcVenice]

I doubt it's as big as the barcelona TLB bug, but I reckon it's newsworthy.

http://www.fudzilla.com/index....iew&id=10707&Itemid=35

 

[dmens]

same as c2d

http://www.realworldtech.com/f...hreadid=80534&roomid=2

 

[VirtualLarry]

Well, it appears that Intel already has a Microcode patch out. So the question is, how much of a performance hit is this fix causing.

 

[Anubis]

WTF does FUD stand for

 

[Idontcare]

Originally posted by: Anubis
WTF does FUD stand for

Fear, uncertainty and doubt

 

[soccerballtux]

Originally posted by: Anubis
WTF does FUD stand for

It stands for "troll".

 

[taltamir]

what destroyed the phenom's sales was that it was expensive, power hungry, and slow... and the TLB was another nail in the coffin.. not because it existed, but because the fix reduced the performance by at least another 10%.

 

[pm]

Originally posted by: Idontcare

Originally posted by: Anubis
WTF does FUD stand for

Fear, uncertainty and doubt

For the longest time I thought it stood for "fscked up data" or words to that effect. Just as for the longest time I thought QFT (Quoted for Truth) meant "Quick Fscking Talking" or words to that effect. Where fsck doesn't mean file system checker as it normally might.

 

[formulav8]

Just like the AMD TLB bug was blown WAY out of proportion, I wouldn't be surprised if some people don't blow Intels TLB bug out of proportion.

Even in truth, The AMD bug was Never duplicated on large scale anyways. Even when AMD had a unofficial contest for people to crash the computer exploiting the TLB issue no one came forward for the prize because no one involved wasn't able to duplicate it in any real world scenarios.

Just my penny on the matter


Jason

 

[BlueBlazer]

Core 2 Duo E7000 and E8000 series http://download.intel.com/desi...or/specupdt/318733.pdf
Exact same text written on page 47.

Core 2 Duo X6800, E6000 and E4000 series http://download.intel.com/desi...or/specupdt/313279.pdf
Exact same text written on page 71.

Pentium Dual Core E2000 series http://download.intel.com/desi...or/specupdt/316982.pdf
Exact same text written on page 59.

Intel Celeron 200 series http://download.intel.com/desi.../specupdt/31854702.pdf
Exact same text written on page 53

There is NO TLB bug in Nehalem http://techreport.com/discussions.x/15979

TECH REPORT:

"While browsing news sites today, we came upon several reports saying Core i7 processors suffer from a translation lookaside buffer-related erratum. Those reports drew parallels between the TLB bug that plagued AMD's first quad-core CPUs and a note in Intel's Core i7 specification update, which states in part:

In rare instances, improper TLB invalidation may result in unpredictable system behavior, such as system hangs or incorrect data. Developers of operating systems should take this documentation into account when designing TLB invalidation algorithms. For the processors affected, Intel has provided a recommended update to system and BIOS vendors to incorporate into their BIOS to resolve this issue.

We asked Intel PR manager Dan Snyder for the chipmaker's official take on the issue, and he replied with the following:

This is simply a pointer to a previous document written in April 2007. This document is an application note (advises on programming techniques) that programmers have had since April of 2007. This item in the Nehalem spec sheet is a web pointer, under the heading "spec clarification". The reporter who wrote this did not contact us and we will try to clarify this with him.

In other words, Intel shareholders probably shouldn't be losing any sleep over this.

For reference, the AMD TLB erratum caused data corruption and system hangs in periods of high CPU utilization, and AMD halted shipments of quad-core Opterons for months because of it. On the desktop, the company released a BIOS fix that crippled the performance of early quad-core Phenoms in many apps. All Phenoms with model numbers ending in "50" and all shipping quad-core Opterons lack the erratum, though.

Update: Snyder has sent us a new statement that makes the situation even clearer:

The "AAJ1 Clarification of TRANSLATION LOOKASIDE BUFFERS" document is a SPEC CLARIFICATION, and is simply a pointer to a previous document written in April 2007.

SPEC CLARIFICATION AAJ1 was initially added due to an issue on the Intel® Core 2 Duo processor which was previously corrected with a BIOS update; this issue does not impact the Nehalem Family of CPUs. There are errata on the Intel® Core i7 processor that relate to the TLB. These all relate to improper translations or error reporting, and all of those that impact functionality have been fixed via BIOS updates prior to Core i7 launch.

Phenom's TLB bug is different type http://techreport.com/discussions.x/13724

In order to better understand this problem, TR spoke with Michael Saucier, Desktop Product Marketing Manager at AMD. Saucier confirmed that the TLB erratum can cause the system to hang when the chip is experiencing high utilization. AMD has stated previously that virtualization workloads can lead to this problem, but Saucier clarified that other workloads can trigger system hangs, as well. He characterized the issue as a race condition in the TLB logic "where the other guy wins who isn't supposed to win," and said the likelihood of the erratum causing a system hang is extremely rare.

Saucier flatly denied any relationship between the TLB erratum and chip clock frequencies. He also said there's no relationship between clock speeds and the performance degradation caused by the BIOS-based fix for the erratum. AMD previously cited the TLB erratum as the primary motivation behind its decision to delay the 2.4GHz Phenom variant.

 

[solofly]

http://theovalich.wordpress.co...is-not-a-story-at-all/

 

[Idontcare]

Originally posted by: pm

Originally posted by: Idontcare

Originally posted by: Anubis
WTF does FUD stand for

Fear, uncertainty and doubt

For the longest time I thought it stood for "fscked up data" or words to that effect. Just as for the longest time I thought QFT (Quoted for Truth) meant "Quick Fscking Talking" or words to that effect. Where fsck doesn't mean file system checker as it normally might.

That's hilarious, and probably is the true origins of the terms. I can see it now, some guy at IBM is telling his coworker that they got nothing to show management at the next project review meeting but a bunch of "fscked up data" which the boss overhears them say and he asks them to repeat it and they said "uh, we mean FUD, you know, uhm, we have some fears and uncertainties and doubts regarding the scientific merit of the data collected thus far".

 

[sivanor]

Originally posted by: BlueBlazer

There is NO TLB bug in Nehalem
http://techreport.com/discussions.x/15979

TECH REPORT:

"While browsing news sites today, we came upon several reports saying Core i7 processors suffer from a translation lookaside buffer-related erratum. Those reports drew parallels between the TLB bug that plagued AMD's first quad-core CPUs and a note in Intel's Core i7 specification update, which states in part:

In rare instances, improper TLB invalidation may result in unpredictable system behavior, such as system hangs or incorrect data. Developers of operating systems should take this documentation into account when designing TLB invalidation algorithms. For the processors affected, Intel has provided a recommended update to system and BIOS vendors to incorporate into their BIOS to resolve this issue.

We asked Intel PR manager Dan Snyder for the chipmaker's official take on the issue, and he replied with the following:

This is simply a pointer to a previous document written in April 2007. This document is an application note (advises on programming techniques) that programmers have had since April of 2007. This item in the Nehalem spec sheet is a web pointer, under the heading "spec clarification". The reporter who wrote this did not contact us and we will try to clarify this with him.

In other words, Intel shareholders probably shouldn't be losing any sleep over this.

For reference, the AMD TLB erratum caused data corruption and system hangs in periods of high CPU utilization, and AMD halted shipments of quad-core Opterons for months because of it. On the desktop, the company released a BIOS fix that crippled the performance of early quad-core Phenoms in many apps. All Phenoms with model numbers ending in "50" and all shipping quad-core Opterons lack the erratum, though.

Update: Snyder has sent us a new statement that makes the situation even clearer:

The "AAJ1 Clarification of TRANSLATION LOOKASIDE BUFFERS" document is a SPEC CLARIFICATION, and is simply a pointer to a previous document written in April 2007.

SPEC CLARIFICATION AAJ1 was initially added due to an issue on the Intel® Core 2 Duo processor which was previously corrected with a BIOS update; this issue does not impact the Nehalem Family of CPUs. There are errata on the Intel® Core i7 processor that relate to the TLB. These all relate to improper translations or error reporting, and all of those that impact functionality have been fixed via BIOS updates prior to Core i7 launch.

[/quote]

I will definitely say that the bug exist. As long as they have to make an Bios patch to correct the processor itself, then there is a flaw.

Also read; improper TLB invalidation may result in unpredictable system

Later revision/stepping of i7 serie hopefully come without this actual failure, so better wait before buying

 

[dmens]

nice bullshit on that link

"Actually, if there are three or more active threads though, the Core i7 processor will not overclock any of its processing cores. Thus, Nehalem is so weak that needs single core acceleration. And that is what helps most of the available applications."

truly a respectable source

 

[Idontcare]

Originally posted by: dmens
nice bullshit on that link

"Actually, if there are three or more active threads though, the Core i7 processor will not overclock any of its processing cores. Thus, Nehalem is so weak that needs single core acceleration. And that is what helps most of the available applications."

truly a respectable source

I haven't seen a strong nehalem to date. At best they don't collapse under their own weight. How many nehalems have you seen accelerate under their own strength? Weakass nehalem, probably can't do a mile under 10 minutes, pissy acceleration to blame I hear.

 

[magreen]

Are we having another use your cpu as twin tailpipes on your car discussion?

@formulav8: LOL, if you use two incorrect double-negatives in one post, do they cancel each other out?

 

[tim924]

It's been discussed before that it's no where near as the good old famous TLB bugs from Phenom,and Intel implied that they had fixed the problem prior to launch and,just in case developers will introduce a bios update which completely remove the bug.Anyways it's really an extremely rare case that you will experience with this TLB bug.It's great to have it in your mind that there is such a thing,but I sensed that some people just want to exerggerate the problem as it's like the phenom,so they could feel better for not upgrading to i7,I know jealousy is part of human instincts,but all the benchmarks have shown that i7 is better than core 2 in terms of multi-threaded programs with a little IPC improvement,so those results dont lie if you are talking about performance-wise.Stability-wise?Haven't really heard anything bad or major problem from the actual users of i7s out there,unless you really want to dig into the little hotter temperature,but i7 is supposed to bear more heat than core 2,so really not a big deal there.

 

[eternalone]

Do you guys think the I7 is the the new Pentium D?????

 

[tim924]

Of course not,but really depends on from what angle you look at it.For instance,if you mainly just use it for gaming ,it's not much of an improvement or shall I say without investing in high-end sli/crossfire.But if you are looking at the original idea of design,since single threaded performance side of processors have maxed out today,multithreaded direction is the only way to further push the performance,if today's games/programs are not designed to take advantage of multithreaded apps,it's actually programmers couldnt keep up with the technology,not i7's improvement are not shown in this case,we just need programmers to do some serious work on their side to utilize i7's massive improvement in hyper-threading.

 

[tim924]

It's just like the E8500 Vs Q6600,gaming experience wise,E8500 would definitely gain some extra frame rates at stock or overclock,but why more people prefer to get the slower clocked 65nm quad?Simply because they know the future is moving toward multi-threaded.For those of you who need better gaming performance,I strongly advice to get a quad core and a better GPU instead of going duo for the 1,or 2 extra fps.

 

[Acanthus]

The only thing stopping me from going i7 is the price of entry.

It is a nice upgrade, but i cant justify the cost.

The Pentium 4-D was a negligible increase across the board, and a flamethrower to boot. That is not the case with i7.

The main issue with the Phenom TLB errata was that there was a significant performance hit or crash associated with real world applications, and that you couldnt fix it for a while. And when they finally did fix it, there was a large performance hit.

 

[aigomorla]

Originally posted by: pm

Originally posted by: Idontcare

Originally posted by: Anubis
WTF does FUD stand for

Fear, uncertainty and doubt

For the longest time I thought it stood for "fscked up data" or words to that effect.

AHAHAHAHAHHAHA your not the only one!

I also thought it ment fubard up data.

 

[Idontcare]

Originally posted by: tim924
Of course not,but really depends on from what angle you look at it.For instance,if you mainly just use it for gaming ,it's not much of an improvement or shall I say without investing in high-end sli/crossfire.But if you are looking at the original idea of design,since single threaded performance side of processors have maxed out today,multithreaded direction is the only way to further push the performance,if today's games/programs are not designed to take advantage of multithreaded apps,it's actually programmers couldnt keep up with the technology,not i7's improvement are not shown in this case,we just need programmers to do some serious work on their side to utilize i7's massive improvement in hyper-threading.

In my current life I spend about 50% of my working time as a programmer. The hardware guys have done a pretty good job of making the performance situation seem like it is the software guys fault for new CPU's not improving the value of the computing experience for the end-user.

The problem is not that programmers are lazy or not smart enough to figure out how to use 8 threads in the applications. The problem (for you the consumer) is that the programmer (nor his employer) is not incentivized to care about it.

When was the last time (if ever) you were faced with the choice of buying two or more software packages that performed the exact same function with the only difference between them being application performance and cost?

Personally I have been in this situation twice, maybe three times. For the rest of my software applications I essentially have no choice, if I needs the app then I buys the software. If it doesn't go fast enough for me then that is my problem as I should have bought the higher GHz processor instead of the cheaper one. No one at Adobe is going to lose their job for not making photoshop an 8-threaded app, and no one is going to specifically not buy photoshop because it isn't an 8-threaded app.

So it isn't really correct to say programmers haven't caught up to the hardware. They aren't bothering to try. Where's the incentive? How many gamers are going to not buy a game if it isn't multi-threaded?

Originally posted by: aigomorla

Originally posted by: pm

Originally posted by: Idontcare

Originally posted by: Anubis
WTF does FUD stand for

Fear, uncertainty and doubt

For the longest time I thought it stood for "fscked up data" or words to that effect.

AHAHAHAHAHHAHA your not the only one!

I also thought it ment fubard up data.

fscked up beyond all recognition up data? That's one fubar'ed acronym :laugh:

 

[VirtualLarry]

Originally posted by: formulav8
Just like the AMD TLB bug was blown WAY out of proportion, I wouldn't be surprised if some people don't blow Intels TLB bug out of proportion.

Even in truth, The AMD bug was Never duplicated on large scale anyways. Even when AMD had a unofficial contest for people to crash the computer exploiting the TLB issue no one came forward for the prize because no one involved wasn't able to duplicate it in any real world scenarios.

Just my penny on the matter


Jason

Not true. It was reported on Linux mailing lists.

 

[VirtualLarry]

Originally posted by: eternalone
Do you guys think the I7 is the the new Pentium D?????

I think it's the new Pentium 60Mhz.